Object Oriented Metrics
Ali Kamandi
Sharif University of Technology
[email protected]
November 2006
1
OO Metrics: Introduction
„
„
„
„
[Pressman]
Measurement and metrics are key components of any
engineering discipline.
The use of metrics for OO systems has progresses much
more slowly than the use of OO methods.
As OO systems become more pervasive, it is essential
that software engineers have quantitative
measurements for assessing the quality of designs at
both the architectural and component levels.
The intent of OO( or non-OO) metrics:
• To better understand the quality of the product.
• To assess the effectiveness of the process.
• To improve the quality of work performed at a
2
project level.
[Pressman 2001]
[RUP]
Principles of Metrics
„
Metrics must be
•
•
•
•
•
„
Simple
Objective
Easy to collect (automated)
Easy to interpret
Hard to misinterpret
The selection of an extensive set of metrics will
depend on the project's characteristics and
context
3
[Pressman]
Metrics for the OO Design Model
„
Whitmire[WHI97] describes nine distinct and measurable
characteristics of an OO design:
• Size (Population, Volume, Length, Functionality)
• Complexity (Interrelated classes)
• Coupling ( No. of Collaborations between classes or messages)
• Sufficiency (the degree to which a design component possesses
features in its abstraction from the current application point of
view)
• Completeness ( same as sufficiency but consider multiple points
of view for fully representation of problem domain)
• Cohesion ( the degree to which the set of properties it possesses
is part of the problem or design domain)
• Primitiveness (the degree to which an operation is atomic)
• Similarity (the degree to which two or more classes are similar)
• Volatility (the degree of change or churn in an artifact because
of defects or changing requirements )
4
Software Metrics
„
„
Some Traditional Metrics
• McCabe Cyclomatic Complexity (CC)
• Lines of Code (LOC)
• Comment Percentage (CP)
Some Object-Oriented Metrics
• Weighted Methods Per Class (WMC)
• Coupling Between Object Classes (CBO)
• Response for a Class (RFC)
• Depth of Inheritance Tree (DIT)
• Lack of Cohesion in Methods (LOCM)
5
Researches on OO Metrics
„
„
„
„
„
„
„
Chidamber and Kemerer: CK Metrics Suite, 1994
CK suite validated by Basili in 1996 and again by Tang in 1999
Many other object-oriented metrics are derived from the CK suite
of object-oriented metrics
Lorenz and Kidd 1994
Harrison, Counsell and Nithi, MOOD Metric Suite, 1998
Whitmire: OO Design Measurements, 1997
…
6
[Pressman]
Class-Oriented Metrics 1
„
Chidamber and Kemerer 1994:
• WMC: weighted methods per class (Σ complexity of methods)
(low)
„ Amount of effort required to implement and test a class
„ Complexity of inheritance tree.
„ Application specific and limiting potential reuse
• DIT: depth of the inheritance tree (Max length node to root)
(low)
„ Lower level classes inherit many methods.
„ Difficulties when attempting to predict the behavior of a class.
„ Greater design complexity.
„ Positive: many methods may be reused
• NOC: number of children ( subclasses that are immediately
subordinate to a class)
7
„ High: reuse, the amount of test
[Pressman]
Class-Oriented Metrics 2
„
Chidamber and Kemerer (Continue):
• CBO: coupling between object classes (No. of coll. in CRC)
(low)
„ High: reusability will decrease, complicated testing and
modifications
„ This is consistent with the general guidline to reduce coupling.
• RFC: response for a class (No. of methods in response set)
(low)
„ High: Testing and overall design complexity will increase.
• LCOM: lack of cohesion in methods (No of methods that
access one or more of the same attributes) (low)
„ High: Complexity of class design, class might be better
designed by breaking it into two or more separate classes.
„ It is desirable to keep cohesion high; that is keep LCOM low.
8
[Pressman]
Class-Oriented Metrics 3
„
„
Lorenz and Kidd [LOR94]:
• Size-oriented: Counts of attributes and operations.
• Inheritance-based: The manner in which operations are
reused through the class hierarchy.
• Internals: Look at cohesion and code-oriented issues.
• Externals: Examine coupling and reuse.
A Sampling of Metrics:
• CS: class size (no. of operations) (no. of attributes) (low)
„ High: class has too much responsibility, reduce the
reusability, complicate implementation and testing
• NOO: number of operations overridden by a subclass (low)
• NOA: number of operations added by a subclass
• SI: specialization index = NOO * level / total no of methods
9
[Pressman]
Class-Oriented Metrics
The MOOD Metrics Suite: Harrison, Counsell and Nithi, 1998
„
„
„
Method inheritance factor: inherited methods / total methods
Coupling factor: formula
Polymorphism factor: formula
10
[Pressman]
Operation-Oriented Metrics
Lorenz and Kidd [LOR94]:
„
„
„
average operation size
operation complexity
average number of parameters per
operation
11
[Pressman]
Testability Metrics
Proposed by Binder [BIN94]:
„
encapsulation related
• lack of cohesion in methods
• percent public and protected
• public access to data members
„
inheritance related
• number of root classes
• fan in: multiple inheritance
• number of children and depth of inheritance tree
12
[Pressman]
OO Project Metrics
Proposed by Lorenz and Kidd [LOR94]:
„
„
„
number
number
number
of
of
of
scenario scripts
key classes
subsystems
13
Req. and Use Case Model Metrics [RUP]
Characteristic
Metrics
Size
Number of requirements
Number of Use Cases
Number of Use Case Packages
Number of scenarios, total and per use case
Number of actors
Length of Use Case (pages of event flow, for example)
Effort
Staff-time units (with production, change and repair separated)
Volatility
Number of defects and change requests (open, closed)
Quality
Defects - number of defects, by severity (open, closed)
Reported complexity (0-5, by analogy with COCOMO [BOE81])
14
Requirements and Use Case Model Metrics (2)
Characteristic
Metrics
Completeness
Use Cases completed (reviewed and under configuration
management with no defects outstanding)/use cases identified (or
estimated number of use cases)
Traceability
Analysis
•
Scenarios realized in analysis model/total scenarios
Design
•
Scenarios realized in design model/total scenarios
Implementation
•
Scenarios realized in implementation model/total
scenarios
Test
•
Scenarios realized in test model (test cases)/total
scenarios
Requirements-to-UC Traceability = Traceable to Use-Case
Model/Total number of requirements
15
OO Design Model Metrics [RUP]
Characteristic
Metrics
Size
Number of classes
Number of subsystems
Number of subsystems of subsystems …
Number of packages
Methods per class, internal, external
Attributes per class, internal, external
Depth of inheritance tree
Number of children
Effort
Staff-time units (with production, change and repair
separated)
Volatility
Number of defects and change requests (open, closed)
16
OO Design Model Metrics (2) [RUP]
Characteristic
Quality
Metrics
Complexity
Response For a Class (RFC): this may be
difficult to calculate because a complete set
of interaction diagrams is needed.
Coupling
Number of children
Coupling between objects (class fan-out)
Cohesion
Number of children
Defects
Number of defects, by severity (open, closed)
Completeness
Number of classes completed/number of classes estimated (identified)
Traceability
Number of classes in Implementation Model/number of classes
17
OO Implementation Metrics (1) [RUP]
Characteristic
Metrics
Size
Number of classes
Number of components
Number of implementation subsystems
Number of subsystems of subsystems …
Number of packages
Methods per class, internal, external
Attributes per class, internal, external
Size of methods, Size of attributes
Depth of inheritance tree
Number of children
Estimated size at completion
Effort
Staff-time units (with production, change and repair separated)
Volatility
Number of defects and change requests (open, closed)
Breakage for each corrective or perfective change, estimated (prior
18
to fix) and actual (upon closure)
OO Implementation Metrics (2) [RUP]
Characteristic
Quality
Completeness
Metrics
Complexity
Response For a Class (RFC)
Cyclomatic complexity of methods*
Coupling
Number of children
Coupling between objects (class fan-out)
Message passing coupling (MPC) [Li and
Henry 1993]
Cohesion
Number of children
Lack of cohesion in methods (LCOM)
Defects
Number of defects, by severity, open, closed
Number of classes unit tested/number of classes in design model
Number of classes integrated/number of classes in design model
Active integration and system test time (accumulated from test
process), that is, time with system operating (used for maturity
calculation)
19
Test Metrics [RUP]
Characteristic
Metrics
Size
Number of Test Cases, Test Procedures, Test Scripts
Effort
Staff-time units for production of test cases, and so on
Volatility
Number of defects and change requests (open, closed)
Quality
Defects — number of defects by severity, open, closed (these are
defects raised against the test model itself)
Completeness
Number of test cases written/number of test cases estimated
Code coverage
Traceability
Number of successful Test Cases in Test Evaluation
Summary/Number of test cases
20
Change Management Metrics [RUP]
Characteristic
Metrics
Size
Number of defects, change requests by severity and status,
also categorized as number of perfective changes, number
of adaptive changes and number of corrective changes.
Effort
Defect repair effort, change implementation effort in stafftime units
Volatility
Breakage (estimated, actual) for the implementation model
subset.
Completeness
Number of defects discovered/number of defects predicted
21
Process Metrics
„
„
„
„
„
„
Duration
Effort
Training time
Inspection time
Meeting time
Process Problem
22
Project Metrics
„
„
„
„
„
Cost/Schedule Control
• BCWS, Budgeted Cost for Work Scheduled
• BCWP, Budgeted Cost for Work Performed
• ACWP, Actual Cost of Work Performed
Modularity = average breakage (NCNB*) per perfective or
corrective change on implementation model
Adaptability = average effort per perfective or corrective
change on implementation model
…
* NCNB is non-comment, non-blank code size.
23
Data Analysis: Example (1)
„
Projects
- Project A: 46 Java Classes (Commercial Software)
- Object-Oriented Constructs
- Project B: 1000 Java Classes (NASA Software)
- Excellent Object-Oriented Constructs
- Project C: 1617 C++ Classes (NASA Software)
- Good Object-Oriented Constructs
24
Data Analysis: Example (2)
„
Correlation of Metrics over Projects
CBOxRFC CBOxWMC RFCxWMC
„
Project
A
0.83
0.43
0.47
Project
B
0.28
0.11
0.75
Project
C
0.35
0.26
0.83
Metric 1 x Metric 2 = correlation between metric
1 and metric 2.
25
Data Analysis: Example (3)
„
Regression Model
- RFC = β1WMC + β2CBO + Constant
„
βWMC
βCBO
Constants
R2
RFCA
0.131
0.777
8.036
0.70
8
RFCB
0.732
0.200
15.427
0.60
8
RFCC
0.792
0.148
6.039
0.70
9
26
References
„
„
„
„
„
„
„
[PRE01] Pressman, R. S., Software Engineering, A Practitioners Approach,
5th Ed., McGraw-Hill, 2001.
[ROY98] Walker Royce 1998. Software Project Management: A Unified
Framework. Addison Wesley Longman.
[CHI94] Chidamber, S. R. and C. F. Kemerer, “A Metrics Suite for Object
Oriented Design”, IEEE Trans. Software Engineering, vol. SE-20, no. 6,
June 1994, pp. 476-493.
[HAR98] Harrison, R. S.J. Counsell, and R. V. Nithi, “An Evaluation of the
MOOD set of Object-Oriented Software Metrics,” IEEE Trans. Software
Engineering, vol. SE-24, no. 6, June 1998, pp. 491-496.
[LOR94] Lorenz, M. and J. Kidd, Object-Oriented Design Metrics, PrenticeHall, 1994.
[WHI97] Whitmire, S., Object-Oriented Design Measurement, Wiley, 1997.
[BIN94] Binder, R. V., “Object-Oriented Software Testing,” CACM, vol. 37,
no. 9, september 1994, p. 29.
27
Questions?
28